Company News · October 31, 2022

Exit Poll Frequently Asked Questions

By edison

Since 2004, The National Election Pool (NEP) and Edison Research have conducted the only national exit polls in the United States. The NEP consists of ABC, CBS, CNN, and NBC, and is the source for projections and analysis for every midterm election, presidential primary, and presidential election.

Below is a list of FAQs about the U.S. election exit polls.

What are election exit polls?
Exit polls are surveys based on interviews with voters as they exit their balloting locations. Exit polls gather voters’ demographic information, feelings on various issues, and for whom they voted. Since not everyone votes in person on Election Day, the exit poll is now a multi-mode survey – exit poll surveys are conducted in person with early voters and Election Day voters and telephone, e-mail and text surveys are conducted with by-mail and early voters.

Why do we need exit polls during U.S. elections?
Voting in the U.S is anonymous. Without surveying voters in some way, only a small amount of information is known about the makeup of the electorate. Without surveys, the only data that would be gathered during an election is the number of votes received by each candidate and eventually which individual voters turned out to vote. Surveys provide insight into how members of different parties, demographic groups, and supporters of particular candidates voted, and why. Exit polls are a dependable way to conduct interviews with actual voters.

How long has Edison Research been conducting exit polls? Edison Research was founded in 1994, so the company has been doing exit polls for 28 years. Edison has a contract with the National Election Pool through 2024, so Edison will have conducted exit polling for over 21 years with the NEP. Since 2017, Edison Research has been reporting the vote count, the exit surveys, and the precinct models for U.S. elections.

What methods are used to conduct exit polls?
Exit polls are conducted via different methods (modes) to reach people who vote in different ways:
1. By-mail voters – reached by a multi-mode telephone, email and text survey of registered voters
2. In-person early voters — interviewers are stationed at early voting centers around the country to conduct in-person surveys with voters right after they cast their ballots
3. In-person on Election Day voters — interviewers are stationed outside polling places on Election Day to conduct in-person surveys with voters right after they cast their ballots

Don’t many people vote early or by mail? Aren’t they missed by your exit polls? In 2020, 70% of the presidential election vote was early or by mail, so we have adjusted our methods to reach all voters regardless of which method they use to cast their vote. Our current expectation is that 40% to 50% of votes in 2022 will be cast before Election Day.

Are exit polls anonymous?
Yes. There is no way to link answers on an exit poll to individual voters. Voters complete their in-person exit poll surveys with paper and pencil with no self-identifying information.

Can exit poll results in the media influence the outcome of the election?
No exit poll data that characterizes the outcome of a race is released until all polls close in that state on election night. Any data you see before election night is NOT exit poll data. Once exit poll data is released when the polls close on election night, it is then up to the media outlet as to how they report the findings.

What about the polls in the days and months leading up to the election?
Those results are from pre-election opinion polls, not exit polls. Many media outlets conduct pre-election polling and many pre-election polls predict which candidate is leading going into an election. Pre-election polls measure how people say they are going upcoming election, but those respondents may or may not end up voting. Exit polls measure behavior and views of actual voters.

How are exit poll responses from Election Day processed so quickly that we see them on election night?  Edison Research hires hundreds of exit poll interviewers across the country to collect the data on Election Day and we have several hundred phone operators taking in the data during Election Day so that we can process the tens of thousands of questionnaires that are filled out by voters quickly for use by our clients on election night.

How long does it take for the NEP to get results from the exit polls?
Edison Research reports the exit poll results in three waves during the day.  The first wave includes all early voter interviews and morning interview on Election Day.  The second wave adds the midday interviews and shortly before poll closing the third wave adds the afternoon and evening interviews.

How long does the exit poll interview last? The exit poll questionnaires typically have about 20 questions and take most voters about five minutes to fill out.

How many people are interviewed for the exit polls?
Edison Research interviewed over 100,000 voters in the 2020 general election. For presidential general elections or midterm elections, Edison Research has people at over 500 Election Day precinct locations. In 2020, we also had interviewers at 200 early voting centers around the country. We interviewed about 30,000 people by telephone before Election Day to identify by mail and early voters.

How many people work on exit polls? Between exit poll interviewers and county vote count reporters, Edison Research will hire over 2,000 people to work for us on Election Day.

Is the “interview” an actual interview?
For voters reached by telephone, it is a telephone interview. For voters reached in person, it is a self-administered paper survey.  For voters reached by email or text it is a self-administered online survey.

Who writes the questionnaire used for exit polls?  A committee of polling directors at the news organizations of the NEP writes the questionnaires, and three of the four members have to agree to put a question into the questionnaire. There is usually consensus. The real issue is that there is a limited amount of space on a self-administered questionnaire. Typically, it is a 5.5 x 8.5 piece of paper and usually fits 18–20 questions.

Aren’t exit polls skewed because some voters are less likely to respond? Edison Research has always measured refusals by gender, age, and race. The response rate for these groups is known and we adjust for it. Warren Mitofsky discovered in the late 1960s that there would be differential response rates by demographic characteristics that interviewers could observe. Age is the main issue. Older people, for example, are less likely to take the exit poll questionnaire. We adjust for that. Over the years, we have found differences by education levels in response rates, with college-educated voters being more likely to fill out the exit poll than non-college-educated voters. That difference has become much more partisan. Democrats have done better with college-educated voters, Republicans with non-college-educated ones. We have evaluated these differences and compared them to the Census surveys, and we have added adjustments. Surveys are going to adapt because no mode is perfect, but by combining multiple modes, you have a better chance of eliminating biases in any one.

Are there ways you can get voters to respond? Using a variety of modes helps combat non-response. Younger people are more likely to respond to text, older people are more likely to respond to telephone, and middle-aged people are more likely to respond to email.

Are you optimistic about the future of the exit poll? (from Joe Lenski’s interview with AEI) “I’m optimistic polling in general still works. As much as our work in 2004, 2016, and 2020 has taken hits for specific errors in specific states and races, the overall average error in surveys is less than it was decades ago. The real issue is educating about the kind of precision you can and can’t demand from these data. The average errors in 1996 were larger than they have been in more recent elections. But Bill Clinton was winning by a landslide, so it didn’t matter. Now we have races decided by 1 or 2 percentage points. Surveys were never designed for that kind of precision. We live in a time where national elections or crucial elections like the Georgia runoff are decided by small margins. There are several things going on. There is a real problem contacting a representative sample. Then there is the trust issue, and then you have really close elections that decide the Electoral College and control of the Senate and House. But we can still tell the story about Trump doing better than expected among Hispanics in certain places. We told the story of the growing cleavage between White college-educated voters and White noncollege-educated voters. We told the story of red states becoming redder and blue becoming bluer. I don’t see 2022 or 2024 being that different.”

Portions of this FAQ were taken from an interview with Edison Research Co-Founder and Executive Vice President Joe Lenski as part of an article from AEI, American Enterprise Institute. For the full interview and more in-depth information about exit polls: https://www.aei.org/wp-content/uploads/2022/01/The-Exit-Polls.pdf?x91208

Get our latest insights delivered to your inbox.